A tensor-based bibliometric framework for measuring research competition across DoD-priority technology domains.
This project came out of a research internship at the Air Force Institute of Technology, working on how the Department of Defense tracks research competitiveness across critical technology domains. Publication counts alone are a weak signal — they say how much a country is publishing, not whether that output is concentrated, accelerating, or shifting between domains. The goal was a framework that could say something about those dynamics directly.
The Department of Defense tracks a defined set of critical technology domains where US-China research competition has direct strategic relevance. A static snapshot of publication share in those domains misses the more useful question: which domains is a given country's research effort moving toward or away from, year over year, and how does that compare to where the other side is moving.
Rather than a flat country-by-domain comparison, I built a three-dimensional tensor — keyword × year × country — from Scopus publication records, so that shifts in research emphasis over time and across domains could be read directly out of the tensor's structure instead of reconstructed from separate cross-tabs.
Structuring the data this way surfaced domain-level trends that a flat publication-count comparison would have flattened out — where research emphasis was concentrating versus spreading, and how that concentration was moving differently across countries and domains over the years covered.
This work was done during a research internship through the Oak Ridge Institute for Science and Education (ORISE) at AFIT, on-site at Wright-Patterson Air Force Base — a different kind of project from the propulsion hardware and modeling work that makes up the rest of this site, but one that sharpened the same instinct: build the framework that actually answers the question, rather than reaching for the nearest existing metric.